[CMake] confused by ctest

[David Cole]

Correct. If you just created your repository today, then it cannot
snap back to a timestamp from yesterday and get a full build tree
based on it. You must have a valid source tree in your repository as
of the most recent nightly start time (in the past) before you can do
a successful Nightly dashboard run.

As you say, .... it *should* just start working tomorrow.

:-)
David


On Tue, Feb 3, 2009 at 6:21 PM, Bill O'Hara <billtohara at gmail.com> wrote:
> Hi David,
>
> Yes, that helps somewhat, thank you! I get it to now use svn and I'm ignore
> the kitware.com trigger mentions as you advise that its not actually doing
> the upload.
>
> Continuous and Experimental cdash submissions seem to go through; I'm
> assuming my nightly one is breaking, where it seems to try to zip back to
> revision 0 and gets confused since there are no files there, because I don't
> have a repository from "yesterday". I assume tomorrow my nightly test will
> automagically begin working if it finds a more recent revision in the right
> time range.
>
>    Old revision of repository is: 5
> * Update repository: "/opt/subversion/bin/svn" info
> D    impl
> D    tests
> D    include
> D    CTestConfig.cmake
> D    CMakeLists.txt
> Updated to revision 0.
> Problem determining the current revision of the repository from output:
> D    impl
> D    tests
> D    include
> D    CTestConfig.cmake
> D    CMakeLists.txt
> Updated to revision 0.
>
> And I fully intend to buy a copy of the book.. once I convince myself with
> some simple tests that its worth dumping time and energy into a full eval of
> cmake on our full hairy source tree. Obvious bootstrapping problem but I
> need to look at a few different systems to narrow a shortlist before
> spending money right now.
>
> thanks again
> b.
>
>
>
> On Tue, Feb 3, 2009 at 12:41 PM, David Cole <david.cole at kitware.com> wrote:
>>
>> Don't worry... it's not submitting results to Kitware, it's using a
>> default value for CTEST_TRIGGER_SITE, which is old and leftover from
>> pre-CDash days... It used to be the case that after all results were
>> uploaded, a "trigger" script would run to tell the server to process the
>> previously uploaded files. CDash processes files as they are submitted so
>> triggers are not necessary with CDash. I actually fixed this very code in
>> ctest earlier today to avoid trigger calls when using CDash. (See changes
>> listed here: http://www.cdash.org/CDash/viewUpdate.php?buildid=263494)
>>
>> To answer your questions:
>> The CMake book has a Tutorial chapter in it that takes you through the
>> steps involved in adding CTest / CDash support to your CMake based project.
>> If you have the book, read through the Tutorial chapter... it's pretty
>> quick. If you don't have the book, buy it. :-)   (You can also check out the
>> code that goes with the chapter in the CMake/Tests/Tutorial/Step1 through
>> Step7 directories in a CMake source tree...)
>> Basically, you need to call ENABLE_TESTING() and INCLUDE(CTest) in your
>> CMakeLists.txt file.
>> And yes, "make test" returns errors when there are test failures...
>> And you should set CTEST_UPDATE_COMMAND to the full path to the svn
>> executable to use svn instead of cvs...
>>
>> HTH,
>> David
>
>